Big Red gets cozy with Google to offer TPUs to AI customers

Oracle Cloud Infrastructure (OCI) customers will soon have a variety of new options, as Oracle expands its partnership with Google and diversifies its chip offerings beyond Nvidia. The company unveiled its plans during an event in Las Vegas this week, highlighting its latest AI investments and introducing Google's Tensor Processing Units (TPUs) to OCI.

This will make Oracle the first hyperscale cloud platform outside of Google's own cloud services to offer TPUs, with availability beginning in 2027 and capacity increasing over several years. Oracle aims to provide customers with greater flexibility to choose the right infrastructure for their AI workloads, based on factors such as performance, throughput, and economics.

Meanwhile, Oracle announced new ChatGPT integrations and a managed platform for cybersecurity that incorporates frontier AI models and Oracle's security expertise. The company also revealed a deal with Snowflake to bring its AI data cloud to OCI as a native service, allowing customers to purchase and use Snowflake directly from Oracle.

Despite Oracle's focus on expanding its chip offerings, the company still announced new Nvidia hardware, including Nvidia Vera Rubin NVL72 systems and Nvidia Vera CPUs, as well as plans for HGX Rubin NVL8 systems. Oracle's record earnings in the most recent quarter, driven by its AI investments, suggest that the company's moves to diversify its offerings may be paying off, but concerns remain about the potential for an AI bubble to burst.